[MPlayer-cvslog] r27849 - trunk/libvo/x11_common.c

[Uoti Urpala]

On Sun, 2008-11-09 at 18:27 +0100, Reimar Döffinger wrote:
> On Sun, Nov 09, 2008 at 05:18:06PM +0200, Uoti Urpala wrote:
> > The link I posted earlier had as good a description of one problem case
> > as any bugreport you're going to get (in short, xserver 1.5.0 implements
> > backing store the same way as composite and it triggers the same
> > problems that non-vsync-aware compositing managers have had).
> 
> No it is not helpful, because now we have reports that it happens _only_
> with the Intel drivers on 1.5 and _not_ with NVidia drivers, that one seems to
> imply it happens with NVidia drivers, but it is most likely fixed with

Wasn't that some separate issue? Intel drivers using blitting instead of
overlay when backing store is enabled (and maybe not specific to 1.5)?

> 1.5.1, so only a bug workaround, so enabling it again without additional
> checks would probably be okay if it ever causes an issue in the future.
> Lastly, in the thread you linked the same issues were reported with the
> 1.4 server and also with OpenGL games.
> So I can't see this as creating anything but more confusion, certainly
> not helpful information.

Those 1.4/OpenGL cases were triggered by composite. Backing store causes
the same issues even when composite is disabled.

> > Also enabling backing store requires more justification than leaving it
> > disabled, and the old change turning it on was no better documented.
> 
> And even more justification is required to apply a patch with a
> completely useless commit message to a completely non-understood issue.
> We can make this change just to try it out and because it seems likely
> to work better (would not be the first time), but a commit message then
> sure must reflect that it is "clueless" code tweaking.
> But lets just stop the pretenses, I just hoped that I could get someone
> who favours this to at least write a good commit message, since they
> obviously know what good it is supposed to do. Doesn't really seem to be
> _that_ important to anyone.

I at least haven't considered fixing this particular commit message to
be that important compared to all the other questionable things in
MPlayer...